A sample is the most convincing object in this trade, and that is exactly the risk. It arrives, it is beautiful, it costs a fraction of an order, and it invites a conclusion much larger than it can support.
The conclusion buyers actually draw is: this is what my order will look like. Sometimes that is true. Often it is not, and the gap is never anyone's dishonesty — it is the difference between what a single object can demonstrate and what a production run has to deliver.
This is about drawing the right conclusion from the right evidence.
The two questions a sample is asked to answer
When a buyer requests a sample, they are usually asking two separate questions without separating them:
"Can this workshop carve to the standard I need?" — a question about craft, inspection and finishing.
"Will my order look like this?" — a question about consistency across many pieces and many blocks.
A sample answers the first question well and the second question poorly. Most disputes in this trade come from a sample being treated as an answer to the second.
What one piece genuinely proves
Everything in this section is real evidence. If a sample fails any of it, stop there.
The carving itself. Outline sharpness, the depth of the eye sockets, whether the jaw line is symmetrical, whether the finish is even across surfaces that face different directions. Craft skill shows up in the places that are hard to reach, not on the front of the face.
The polish. On dense material this is where a workshop's standard becomes visible. Choosing stone for carved skulls covers how different materials respond, but the short version is that a poor polish cannot be corrected later, and it is obvious in the hand even when it photographs well.
The natural flaws left in. Every piece has some — a small pit, a mineral line, a slight asymmetry where the block dictated the form. What matters is whether they were disclosed before you saw the piece. Natural flaws versus defects draws the line. A supplier who points them out unprompted is telling you how they will behave at volume.
The disclosure standard. Does the piece arrive with its material stated by mineral name, its treatment stated explicitly, and its weight and dimensions recorded? A sample with no paperwork is a sample from a workshop that does not keep records, and you will want records later.
That the workshop exists. A sample that appears quickly from nowhere, with no traceable origin, proves nothing about who will make the rest. Origin and maker are the fields to check — the same ones a spec sheet has to carry.
What one piece cannot prove
This is the part that gets skipped, and it is the expensive part.
It cannot prove colour will repeat. Natural stone does not repeat. Colour, banding, veining and internal structure differ from block to block, and the next block is already in the yard. If you approve a sample and then order two hundred pieces, the run will not be identical to the sample, and no workshop can make it so. What a good supplier can do is tell you the realistic spread in advance and show you pieces from more than one block.
It cannot prove weight and dimension consistency. Hand-carved pieces vary. A sample weighing one figure tells you the order of magnitude, not the weight of every unit. This is why per-piece data matters more than any single sample.
It cannot prove packing survives transit. One piece, packed for one shipment, is a weak test of how a pallet behaves. Packing is a separate question with its own answer.
It cannot prove lead time. A sample is often made ahead of the queue precisely because it is small. It is not a measurement of how long a production run will take.
It cannot prove the price holds at volume. Unit cost at sample quantity and unit cost at production quantity are different numbers, for reasons that have nothing to do with goodwill.
The trap: approving the sample, not the specification
There is a specific failure that is worth naming, because it is common and it is expensive.
A sample arrives. It is good. The buyer approves it mentally — and the order proceeds against nothing written down. Six weeks later the pieces arrive with a finish the buyer considers slightly softer than the sample, and there is no document to appeal to, because the agreement was a feeling about an object.
The fix is procedural, not adversarial: approve a specification, using the sample as one exhibit among several. Concretely, before production starts, put in writing:
- The material by mineral name, and the treatment status
- The dimension tolerances you will accept
- The finish standard, described in words, with the sample retained as the reference object — gloss, satin and matte fail differently, so naming the surface matters as much as naming the material (polished, satin and matte)
- The realistic colour and pattern spread, ideally illustrated with pieces from more than one block
- What happens when a piece falls outside those tolerances
That list is not distrust. It is the only way a sample can be reused as a reference months later, when memories of it have drifted.
> Keep the approved sample. If it goes back into general stock, or is sold, you have lost your own reference point. A retained sample is the cheapest piece of insurance in this trade.
Paid sampling versus a stock first order
These are two different transactions and it helps to keep them apart.
Paid sampling applies to custom and OEM work, where something genuinely has to be made to your specification — a new form, a size outside the stock range, your own mark on the piece. Nothing exists yet, so something has to be made, and that work has a cost. Our position on custom sampling, and how the sample relates to the order that follows, is set out in ordering custom crystal skulls.
A stock first order is the better route when you simply want to see the quality. Rather than paying for one object you cannot sell, you buy a small set of real pieces from the current range — pieces that go straight into your store. This is our normal recommendation, and the reasoning is in the MOQ guide rather than repeated here.
The distinction matters for a practical reason: a sample you commissioned answers questions about a form that does not exist yet. A stock piece answers questions about the pieces you will actually be reselling.
How to review a sample properly
A sample review that lasts four minutes is not a review. What to actually do, in order:
- Weigh it on a kitchen scale and compare against the recorded weight. Density varies between materials and even within one, so a large deviation is worth a question.
- Measure it in the dimension convention you will use for the order — length is not enough for a piece that is not a cube.
- Turn it over. The base, the underside of the jaw, the inside of the eye sockets. Craft standards show where the light does not.
- Find the flaws deliberately, then check whether they were disclosed. Undisclosed flaws are a process problem; disclosed ones are just stone.
- Photograph it in the conditions you will sell it in. Retail lighting is not studio lighting, and a piece that only works under controlled light will disappoint you on a shelf.
- Ask what would differ in a production run, in the supplier's own words. A supplier who says "nothing" is either new to this or hoping you are.
- Write down what you approved before you approve it.
Item six is the one most worth asking for, because it costs the supplier something to answer honestly.
A short reading list of realistic expectations
- Finish: expected to hold closely, because it comes from the same hands and the same abrasives
- Outline and proportion: expected to hold closely within stated tolerances
- Weight: expected to vary, because density varies
- Colour and pattern: expected to vary meaningfully, often more than buyers expect
- Presence of small natural features: expected, and the amount is not controllable in advance
If a supplier's sample conversation treats the last three the same way as the first two, the conversation is optimistic rather than informed.
How we handle it
Our range is 18 materials across 20 stock designs, spanning 1.32 lb (600 g) to 9.08 lb (4120 g) and 3.8″ to 7.1″ in length, with weight and dimensions recorded per piece rather than averaged. Every piece in the current range is untreated, and we state that explicitly. Because a single material can carry more than one design — Rose Quartz and Moss Agate each appear in two figures — we point out when a photograph shows a different design in the same material rather than letting the comparison carry.
Each finished piece is inspected individually by hand before it enters inventory, in our own workshop in Donghai, Jiangsu — approximately 300 m², 15 machines, a team of more than 20. That inspection standard is what a sample is really testing.
If you want to see the actual quality, request a quote for a small set of stock pieces. If you have a form in mind that we do not make, the custom process explains how a paid sample works and how it relates to production. Availability, quantity, pricing, payment terms, freight and lead time are confirmed in our written quotation — nothing is committed until both sides confirm in writing.
The short version
- A sample tests craft well and consistency poorly. Separate the two.
- It cannot prove colour repeat, weight consistency, packing performance or lead time.
- Approve a written specification, with the sample as one exhibit — not the sample alone.
- Keep the approved sample. It is your only reference point later.
- Paid sampling is for custom work that must be made; for stock quality, a small set of real pieces is the better test.
- Ask the supplier what will differ at volume, in their own words. The answer tells you a lot.